Am 20.12.2018 um 11:54 schrieb Riccardo (Jack) Lucchetti:
On Thu, 20 Dec 2018, Sven Schreiber wrote:
> I'll admit right away that this question has no real practical problem
> behind it, it's just out of internal consistency:
> The mean() function accepts a list and then averages across the
> members of the list (and returns a series of per-obs-averages).
> In contrast, the pmean() function doesn't have this second use, it
> only accepts one series (as does mean). Any particular reason why not?
> The same goes for pmax(), pmin(), psd(), psum(), and perhaps also
> pxsum(), I guess.
Well, this makes sense I suppose, snce pmean() and friends are supposed
to operate "vertically", so to speak, through time, so I can't really
imagine what you might want to have with a list argument. Perhaps
Well the standard mean() function also changes the direction in which it
operates, this is exactly the trick when you give it a list.
But you're right that mean() also already works in a panel dataset (I
hope), and so doing pmean(mean(X)) should be all one needs.